I'm building an open-source robot trader framework.

Discussion in 'App Development' started by jbax, Dec 2, 2019.

  1. jbax

    jbax

    My dude this is full open source. I have no control whatsoever. You can copy, modify, sell, do whatever you want. There is no dependency on any third party server/setup/etc. It all depends on the data you collect yourself (the framework helps you to manage it), and the code you write. You can change how it will execute orders (if at all, you might just want it to send you a notification so you can review and trade manually).
     
    #11     Dec 3, 2019
  2. traider

    traider

    have u looked at pylivetrader and quantopian?
    There is also backtrader , quantconnect
     
    #12     Dec 3, 2019
    sharpquant likes this.
  3. jbax

    jbax

    Some of them yes. quantopian/quantconnect depend on their platform so you have limited control over what you are doing.

    pylivetrader if I am not mistaken loads all data in memory so it becomes hard to backest efficiently, especially if you are trying to emulate a live market situation (trading multiple symbols at the same time, using different time scales)

    backtrader is GPL3 which I'm personally iffy about. Anyway that one is pretty complete but complex to use and yet I could not find how to cleanly manage open orders (for example, when to stop a trade before one or more strategies give a sell signal) and optimize parameters efficiently (especially if you have thousands of combinations to test).
     
    Last edited: Dec 3, 2019
    #13     Dec 3, 2019
    sharpquant likes this.
  4. traider

    traider

    quantopian has open source version called zipline
    quantconnect has lean
     
    #14     Dec 3, 2019
  5. d08

    d08

    How does backtrader do for speed compared to using pandas/numpy? It seems to be using just python to loop over bars, that sounds slow.
    I'm aware a vectorized pandas solution is a bit more limited when it comes to certain strategies though. This is what I'm dealing with now and it's quite a puzzle. I'm thinking of a two variant approach where you have the "raw" simulation which is purely vectorized for speed and a "realistic" portfolio simulation based on a equity curve.
     
    #15     Dec 3, 2019
  6. d08

    d08

    So let me get this straight. You'd recommend a closed source project like NinjaTrader which really can do anything without you knowing (not saying it does) vs an open source project where all your code is right in front of you to digest and modify?
    I hope you realize the absurdity.
     
    #16     Dec 3, 2019
  7. traider

    traider

    I'm curious what happens if ninjatrader has a bug and you lose all your money because of it?
     
    #17     Dec 3, 2019
  8. d08

    d08

    I'm pretty certain the agreement you accept has it covered and they are not liable. I wouldn't expect them to be.
     
    #18     Dec 3, 2019
  9. I wouldn't suggest backtrader to you because you're a DIYer. But BT has a lot of great ideas on how to structure API's.
     
    #19     Dec 3, 2019
  10. IAS_LLC

    IAS_LLC

    Java. Eww.
     
    #20     Dec 3, 2019
    nooby_mcnoob likes this.